tvl-depot/users/wpcarro/scratch/deepmind/part_two/word-cloud.py

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

80 lines
2.1 KiB
Python
Raw Normal View History

import unittest
import re
from collections import Counter
class WordCloudData(object):
def __init__(self, x):
x = x.replace('...', ' ').replace(' - ', ' ')
x = ''.join(c for c in x if c not in ',.!?;:')
self.words_to_counts = dict(
Counter(x.lower() for x in re.split(r'\s+', x)))
# Tests
class Test(unittest.TestCase):
def test_simple_sentence(self):
input = 'I like cake'
word_cloud = WordCloudData(input)
actual = word_cloud.words_to_counts
expected = {'i': 1, 'like': 1, 'cake': 1}
self.assertEqual(actual, expected)
def test_longer_sentence(self):
input = 'Chocolate cake for dinner and pound cake for dessert'
word_cloud = WordCloudData(input)
actual = word_cloud.words_to_counts
expected = {
'and': 1,
'pound': 1,
'for': 2,
'dessert': 1,
'chocolate': 1,
'dinner': 1,
'cake': 2,
}
self.assertEqual(actual, expected)
def test_punctuation(self):
input = 'Strawberry short cake? Yum!'
word_cloud = WordCloudData(input)
actual = word_cloud.words_to_counts
expected = {'cake': 1, 'strawberry': 1, 'short': 1, 'yum': 1}
self.assertEqual(actual, expected)
def test_hyphenated_words(self):
input = 'Dessert - mille-feuille cake'
word_cloud = WordCloudData(input)
actual = word_cloud.words_to_counts
expected = {'cake': 1, 'dessert': 1, 'mille-feuille': 1}
self.assertEqual(actual, expected)
def test_ellipses_between_words(self):
input = 'Mmm...mmm...decisions...decisions'
word_cloud = WordCloudData(input)
actual = word_cloud.words_to_counts
expected = {'mmm': 2, 'decisions': 2}
self.assertEqual(actual, expected)
def test_apostrophes(self):
input = "Allie's Bakery: Sasha's Cakes"
word_cloud = WordCloudData(input)
actual = word_cloud.words_to_counts
expected = {"bakery": 1, "cakes": 1, "allie's": 1, "sasha's": 1}
self.assertEqual(actual, expected)
unittest.main(verbosity=2)